[freenet-dev] Infocalypse feedback (Mercurial over Freenet)
On Monday, 11. May 2009 21:20:49 Arne Babenhauserheide wrote: > The only problem I still have is that keeping the uris in the central > config file didn't work (all paths in the config file were lowercase while > the real paths aren't - maybe that's connected to the issue). This problem seems fixed by now; works for me on amd64, Gentoo GNU/Linux - thanks to feral code wright! Information: $ hg version Mercurial Distributed SCM (version 437e06bbd11e+20090515) # means: current hg head from the main development repo $ python --version Python 2.5.4 Best wishes, Arne --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- - singing a part of the history of free software - http://infinite-hands.draketo.de -- next part -- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part. URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090516/8e93f3d1/attachment.pgp>
[freenet-dev] Usability test results
On Saturday 16 May 2009 15:02:19 Thomas Sachau wrote: > Matthew Toseland schrieb: > > On Friday 15 May 2009 16:35:40 Thomas Sachau wrote: > >> Matthew Toseland schrieb: > >>> On Thursday 14 May 2009 18:35:07 Thomas Sachau wrote: > >>>> Matthew Toseland schrieb: > >>>>> My observation: Can we get rid of the "I will configure it manually" > >>>>> choice? > >>>>> And maybe the welcome page? (#3094) > >>>> You want to force everyone to use the Wizard? > >>> Why would that be bad? > >> What if i dont want to do use the Wizard? Also, if i removed the "wizard > > done" line (intentinally or > >> by mistake), a new run would remove my custom settings. With the option, i > > can just stop the wizard > >> and no harm done. > > > > If you know enough to skip the wizard you should shutdown the node, edit the > > config file and tell the node you have done the wizard! > > Is there a need for editing the config file? You can set everything with the config section too, but > without the "i want to do it myself", you cant disable the wizard from the GUI. > > >>>>> Related idea: We should maybe tell the user in the installer that they > >>>>> should > >>>>> use a separate browser for Freenet, rather than in the wizard? And then > >>>>> let > >>>>> them choose one, and then use it when they click on the icon to browse > >>>>> Freenet? (#3104) > >>>> This would produce additional work for people packaging freenet, since > > they > >>>> would have to warn the > >>>> user themselves, while users tend to ignore the output of the package > >>>> manager. > >>>> So this would lower the chance of people noticing the request for a > >>>> different freenet > >>>> browser/profile and therefor i am against it. I suggest the current way: > >>>> Warning during first call > >>>> of the webinterface like it is currently done. > >>> Well, maybe on linux, with the packages that we don't have yet... > >> Did you miss the Gentoo ebuilds? > >> Isnt it a goal to get other distros to package it too? Just because it did > > not happen until now, > >> doesnt mean it wont happen some time in the future. May just need more time > > since Gentoo as source > >> based distro may be a bit better for packages than binary distros. > > > > No, it is a goal to package it with private repositories. Having a debian > > package that is frozen for 3 years is not useful at the present time. > >> And if we have it for linux, why would you like to add additional code for > > windows (both in the > >> installer and in freenet, which would have to detect the OS and then decide > > to show the warning or > >> not)? > > > > Well, we could do something similar for *nix, no? Launch a suitable privacy > > enabled browser when the user runs the browse-freenet script? > > You dont know the user system. While windows user systems may be similar to each others, this is not > true for linux. Where would you place that script? How would you check which browser the user wants > to use? This idea looks more like the way user handling is done on windows or ubuntu: Expect him to > know nothing and try to do everything for him. Might be nice for beginners and if it works, but > makes things worse for experienced users, who want to do it different and also makes it harder, if > there are problems. > Imho you cant beat stupidity. Either users read a message and act the right way or they dont. You > cannot prevent them from doing bad things. > > Additionally, Gentoo is about choice, if there is a warning, the user can choose, with a forcing > script, there is no choice, which is a bad idea for this philosophy, therefor i vote against such a > script for linux. Well, we already have a Browse Freenet script on all three platforms. Currently it detects browsers that we know about. You don't have to use it if you don't want to. But we should extend it to use incognito mode if possible, and to favour browsers with such support. I dunno how we can determine whether such a mode works with the particular installed version though... -- next part -- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 835 bytes Desc: This is a digitally signed message part. URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090516/9e27cd6f/attachment.pgp>
[freenet-dev] Please help us test the new wininstaller (Vista users especially welcome)
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Using the current wininstaller (FreenetInstaller-1210), I installed into Windows 7 x64 Build 7100. Installed java as prompted (x32). Install went very well. The uninstaller leaves user 'Freenet' and I updated my open ticket(bug) with these results. Freenet found that I had Chrome installed and used it, even though it was not my default browser.Bug/Feature? -BEGIN PGP SIGNATURE- Charset: UTF8 Note: This signature can be verified at https://www.hushtools.com/verify Version: Hush 3.0 wkYEARECAAYFAkoPUi8ACgkQDAg0OvA3V4BvTACfctkSvN531D0WN5mDSyFd1bcTCHMA n0Fmqx36EEj+tFzptUzTxsAYgtwX =Wrc+ -END PGP SIGNATURE- -- Free information on a POS system that meets your needs. Click here. http://tagline.hushmail.com/fc/BLSrjkqZugEikWAyHtCFVznVXfLKxokMTdj0gKb01Zq8iQxm3rynr2M6HnC/
[freenet-dev] Usability test results
Matthew Toseland schrieb: > On Friday 15 May 2009 16:35:40 Thomas Sachau wrote: >> Matthew Toseland schrieb: >>> On Thursday 14 May 2009 18:35:07 Thomas Sachau wrote: >>>> Matthew Toseland schrieb: >>>>> My observation: Can we get rid of the "I will configure it manually" >>>>> choice? >>>>> And maybe the welcome page? (#3094) >>>> You want to force everyone to use the Wizard? >>> Why would that be bad? >> What if i dont want to do use the Wizard? Also, if i removed the "wizard > done" line (intentinally or >> by mistake), a new run would remove my custom settings. With the option, i > can just stop the wizard >> and no harm done. > > If you know enough to skip the wizard you should shutdown the node, edit the > config file and tell the node you have done the wizard! Is there a need for editing the config file? You can set everything with the config section too, but without the "i want to do it myself", you cant disable the wizard from the GUI. >>>>> Related idea: We should maybe tell the user in the installer that they >>>>> should >>>>> use a separate browser for Freenet, rather than in the wizard? And then >>>>> let >>>>> them choose one, and then use it when they click on the icon to browse >>>>> Freenet? (#3104) >>>> This would produce additional work for people packaging freenet, since > they >>>> would have to warn the >>>> user themselves, while users tend to ignore the output of the package >>>> manager. >>>> So this would lower the chance of people noticing the request for a >>>> different freenet >>>> browser/profile and therefor i am against it. I suggest the current way: >>>> Warning during first call >>>> of the webinterface like it is currently done. >>> Well, maybe on linux, with the packages that we don't have yet... >> Did you miss the Gentoo ebuilds? >> Isnt it a goal to get other distros to package it too? Just because it did > not happen until now, >> doesnt mean it wont happen some time in the future. May just need more time > since Gentoo as source >> based distro may be a bit better for packages than binary distros. > > No, it is a goal to package it with private repositories. Having a debian > package that is frozen for 3 years is not useful at the present time. >> And if we have it for linux, why would you like to add additional code for > windows (both in the >> installer and in freenet, which would have to detect the OS and then decide > to show the warning or >> not)? > > Well, we could do something similar for *nix, no? Launch a suitable privacy > enabled browser when the user runs the browse-freenet script? You dont know the user system. While windows user systems may be similar to each others, this is not true for linux. Where would you place that script? How would you check which browser the user wants to use? This idea looks more like the way user handling is done on windows or ubuntu: Expect him to know nothing and try to do everything for him. Might be nice for beginners and if it works, but makes things worse for experienced users, who want to do it different and also makes it harder, if there are problems. Imho you cant beat stupidity. Either users read a message and act the right way or they dont. You cannot prevent them from doing bad things. Additionally, Gentoo is about choice, if there is a warning, the user can choose, with a forcing script, there is no choice, which is a bad idea for this philosophy, therefor i vote against such a script for linux. -- next part -- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 315 bytes Desc: OpenPGP digital signature URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090516/4f2da134/attachment.pgp>
[freenet-dev] Usability test results
* xor [2009-05-15 22:07:34]: > On Friday 15 May 2009 20:04:58 Robert Hailey wrote: > > On May 15, 2009, at 11:57 AM, xor wrote: > > > On Friday 15 May 2009 18:37:12 Thomas Sachau wrote: > > >> xor schrieb: > > >>> On Thursday 14 May 2009 19:35:07 Thomas Sachau wrote: > > >>>> Matthew Toseland schrieb: > > >>>>> My observation: Can we get rid of the "I will configure it > > >>>>> manually" > > >>>>> choice? And maybe the welcome page? (#3094) > > >>>> > > >>>> You want to force everyone to use the Wizard? > > >>>> > > >>>>> Because we were both on the same LAN, it did not connect, until > > >>>>> I told > > >>>>> him to set it to allow local addresses on that peer. There > > >>>>> should be a > > >>>>> checkbox when adding a noderef, defaulting to on, "Friend may be > > >>>>> on the > > >>>>> same local network as me" or something. (#3098) > > >>>> > > >>>> This is imho not usual, so i would set this to very low priority > > >>>> and > > >>>> only for advanced mode enabled. > > >>> > > >>> IMHO the current behavior is a "bug" though: We have the MDNS > > >>> plugin, > > >>> which is installed by default and should automatically (opennet-) > > >>> connect > > >>> nodes on a LAN. However it does not work because LAN connections are > > >>> disallowed by default. > > >> > > >> Afaik that plugin is no longer installed by default. > > > > > > Why not? I like the idea of self-organizing Freenet on LANs very > > > much :) > > > > > > Especially if you consider large public or private LANs where two > > > people > > > running Freenet might not even know each others, so they couldn't > > > get manually > > > connected. > > > > Hmmm... but does a node want to advertise that we are running freenet? > > Maybe dependent on the security level setting. > > The more people confess that they are running Freenet, the more difficult it > becomes to blame an individual as guilty for running it. > > Further, if you run opennet, you are screwed anyway, and MDNS is mostly > useful > for Opennet. > > > In either case, a small network would have major issues if it was not > > also connected to the large network, and so I think still that > > automatically detecting locality based on external ip address would > > help (e.g. a node could preferentially keep any opennet peers that it > > finds as being local). > > Wouldn't it take much load off the "internet", i.e. small bandwidth > connections, if any nodes which are connected via LAN used the LAN for > routing > requests if possible? > > I think the LAN could behave like a "single large node", i.e. requests for > data which is stored within the LAN would be answered very fast, and much > data > could be stored if there are many nodes on a LAN. > The idea is not new, at the beginning bloom-filter sharing was supposed to be for darknet nodes or "local" nodes. There is also a ticket on mantis describing that -- next part -- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 197 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090516/cd49a389/attachment.pgp>
[freenet-dev] Usability test results
* Robert Hailey [2009-05-15 13:04:58]: > > On May 15, 2009, at 11:57 AM, xor wrote: > > > On Friday 15 May 2009 18:37:12 Thomas Sachau wrote: > >> xor schrieb: > >>> On Thursday 14 May 2009 19:35:07 Thomas Sachau wrote: > >>>> Matthew Toseland schrieb: > >>>>> My observation: Can we get rid of the "I will configure it > >>>>> manually" > >>>>> choice? And maybe the welcome page? (#3094) > >>>> > >>>> You want to force everyone to use the Wizard? > >>>> > >>>>> Because we were both on the same LAN, it did not connect, until > >>>>> I told > >>>>> him to set it to allow local addresses on that peer. There > >>>>> should be a > >>>>> checkbox when adding a noderef, defaulting to on, "Friend may be > >>>>> on the > >>>>> same local network as me" or something. (#3098) > >>>> > >>>> This is imho not usual, so i would set this to very low priority > >>>> and > >>>> only for advanced mode enabled. > >>> > >>> IMHO the current behavior is a "bug" though: We have the MDNS > >>> plugin, > >>> which is installed by default and should automatically (opennet-) > >>> connect > >>> nodes on a LAN. However it does not work because LAN connections are > >>> disallowed by default. > >> > >> Afaik that plugin is no longer installed by default. > > > > Why not? I like the idea of self-organizing Freenet on LANs very > > much :) > > > > Especially if you consider large public or private LANs where two > > people > > running Freenet might not even know each others, so they couldn't > > get manually > > connected. > > > > Hmmm... but does a node want to advertise that we are running freenet? > Maybe dependent on the security level setting. > > In either case, a small network would have major issues if it was not > also connected to the large network, Ever heard of darknet? > and so I think still that > automatically detecting locality based on external ip address would > help (e.g. a node could preferentially keep any opennet peers that it > finds as being local). > That's what darknet connections are about. Opennet nodes connecting to the same darknet nodes is a BUG which has already been reported and has a ticket on mantis. -- next part -- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 197 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090516/9c32cc4f/attachment.pgp>
[freenet-dev] Usability test results
* xor [2009-05-15 18:24:53]: > On Thursday 14 May 2009 19:35:07 Thomas Sachau wrote: > > Matthew Toseland schrieb: > > > My observation: Can we get rid of the "I will configure it manually" > > > choice? And maybe the welcome page? (#3094) > > > > You want to force everyone to use the Wizard? > > > > > Because we were both on the same LAN, it did not connect, until I told > > > him to set it to allow local addresses on that peer. There should be a > > > checkbox when adding a noderef, defaulting to on, "Friend may be on the > > > same local network as me" or something. (#3098) > > > > This is imho not usual, so i would set this to very low priority and only > > for advanced mode enabled. > > > > IMHO the current behavior is a "bug" though: We have the MDNS plugin, which > is > installed by default and should automatically (opennet-) connect nodes on a > LAN. Did you even read the code? It was never intended to do that and doesn't. > However it does not work because LAN connections are disallowed by > default. > IT WASN'T IMPLEMENTED AT ALL. -- next part -- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 197 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090516/bfcadeac/attachment.pgp>
[freenet-dev] What should be in 0.8, and when should it be released?
On Saturday 16 May 2009 10:12:06 Luke771 wrote: > Matthew Toseland wrote: > > Ian has found some more funding, so we are okay for another 6-8 months. This > > will be announced properly shortly. This relieves the pressure to get > > something out regardless of what it is. So we can have a sensible discussion > > on: > > 1. What should be in 0.8? > > 2. When to release 0.8? > > > > My current view: > > > > We have a pretty good basis for a release already, but we need a lot more > > debugging, we probably need a load more optimisation on db4o (if discussions > > on the support list are anything to go by), there are a few features we have > > already agreed are essential, which are mostly easy, there is Freetalk, and > > there is Bloom filter sharing. IMHO Bloom filter sharing will make a *BIG* > > difference to Freenet's performance and therefore to the success (users, > > content, donations) of 0.8.0. > > > > We should aim for a beta in the not too distant future, however IMHO we should > > aim to implement the essential and very important features below before > > releasing it, and even a beta needs to be reasonably usable in terms of bugs; > > we will very likely need further optimisation on db4o to make Freenet > > practically usable for all users. > > > > So I am suggesting that we allow 2-3 months for feature work, fixing major > > bugs, and further optimising db4o, and release a beta in July. After that we > > have final bugfixing aiming for a final 0.8.0 before say October. > > > > FEATURES: > > > > Essential: > > - Freetalk. Freetalk needs to work well. Any essential plugin changes for > > Freetalk. (Mostly down to xor) > > - The new wininstaller. This is essential, we need Vista support, > > unfortunately it's not going to disappear much as we might want it to! (Easy) > > - MHKs (or DHKs) - redundant CHKs for the top block of a splitfile (no, I > > don't like RHK, it sounds too much like RSK, and R stands for Revocable not > > Redundant). (Reasonably easy) > > - Various minor usability tweaks. (Mostly easy but lots of it) > > - Fix, or at least disappear, the "N peers forcibly disconnected due to not > > acknowledging packets" bug/message. (???) > > > > Very important: > > - Bloom filter sharing and all its dependancies: Client-cache, don't cache > > while HTL is at maximum, transitional arrangements for old nodes' datastores. > > (Estimate 4 weeks for working and thoroughly debugged) > > > > Fairly important: > > - Multi-container freesite inserts. saces' work is pretty much working afaik, > > but it needs to be made fully persistence aware and debugged and de-leaked. > > - Update plugins over Freenet. > > - Pause mode. > > > > Easy and would be nice: > > - Solve the last-segment-has-less-redundancy problem for splitfiles. > > - Guesstimate the monthly transfer usage for each bandwidth setting on the > > bandwidth selection screen. > > - Deploy the new x86-64 FEC code. > > - IM-friendly noderefs. > > > > Would be nice: > > - Javascript support for loading inline images on a page. > > - Audio/video filters, plus an increased maximum file size when using the > > progress page. > > - Hierarchical DBRs for USK updating. > > > Loading official plugins off Freenet using the drop-down menu would be nice. > I understand that this is not going to be in 0.8 because it requires > official repositories on freenet with revocation key and everything, but > IMHO it should have at least 'some' priority after 0.8 is released. Well, maybe for 0.8. The immediate future is 0.7.5. > > Also, would it be possible to make an alternate update script that > fetches the latest .jar off freenet instead of the website using that > CLI tool mentioned in config page? That would require a working node, in which case update over Freenet should work. -- next part -- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 835 bytes Desc: This is a digitally signed message part. URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090516/2895088b/attachment.pgp>
[freenet-dev] Usability test results
On Friday 15 May 2009 16:35:40 Thomas Sachau wrote: > Matthew Toseland schrieb: > > On Thursday 14 May 2009 18:35:07 Thomas Sachau wrote: > >> Matthew Toseland schrieb: > >>> My observation: Can we get rid of the "I will configure it manually" > >>> choice? > >>> And maybe the welcome page? (#3094) > >> You want to force everyone to use the Wizard? > > > > Why would that be bad? > > What if i dont want to do use the Wizard? Also, if i removed the "wizard done" line (intentinally or > by mistake), a new run would remove my custom settings. With the option, i can just stop the wizard > and no harm done. If you know enough to skip the wizard you should shutdown the node, edit the config file and tell the node you have done the wizard! > > >>> Related idea: We should maybe tell the user in the installer that they > >>> should > >>> use a separate browser for Freenet, rather than in the wizard? And then > >>> let > >>> them choose one, and then use it when they click on the icon to browse > >>> Freenet? (#3104) > >> This would produce additional work for people packaging freenet, since they > >> would have to warn the > >> user themselves, while users tend to ignore the output of the package > >> manager. > >> So this would lower the chance of people noticing the request for a > >> different freenet > >> browser/profile and therefor i am against it. I suggest the current way: > >> Warning during first call > >> of the webinterface like it is currently done. > > > > Well, maybe on linux, with the packages that we don't have yet... > > Did you miss the Gentoo ebuilds? > Isnt it a goal to get other distros to package it too? Just because it did not happen until now, > doesnt mean it wont happen some time in the future. May just need more time since Gentoo as source > based distro may be a bit better for packages than binary distros. No, it is a goal to package it with private repositories. Having a debian package that is frozen for 3 years is not useful at the present time. > > And if we have it for linux, why would you like to add additional code for windows (both in the > installer and in freenet, which would have to detect the OS and then decide to show the warning or > not)? Well, we could do something similar for *nix, no? Launch a suitable privacy enabled browser when the user runs the browse-freenet script? -- next part -- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 835 bytes Desc: This is a digitally signed message part. URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090516/acfb23dc/attachment.pgp>
[freenet-dev] What should be in 0.8, and when should it be released?
Matthew Toseland wrote: > Ian has found some more funding, so we are okay for another 6-8 months. This > will be announced properly shortly. This relieves the pressure to get > something out regardless of what it is. So we can have a sensible discussion > on: > 1. What should be in 0.8? > 2. When to release 0.8? > > My current view: > > We have a pretty good basis for a release already, but we need a lot more > debugging, we probably need a load more optimisation on db4o (if discussions > on the support list are anything to go by), there are a few features we have > already agreed are essential, which are mostly easy, there is Freetalk, and > there is Bloom filter sharing. IMHO Bloom filter sharing will make a *BIG* > difference to Freenet's performance and therefore to the success (users, > content, donations) of 0.8.0. > > We should aim for a beta in the not too distant future, however IMHO we > should > aim to implement the essential and very important features below before > releasing it, and even a beta needs to be reasonably usable in terms of bugs; > we will very likely need further optimisation on db4o to make Freenet > practically usable for all users. > > So I am suggesting that we allow 2-3 months for feature work, fixing major > bugs, and further optimising db4o, and release a beta in July. After that we > have final bugfixing aiming for a final 0.8.0 before say October. > > FEATURES: > > Essential: > - Freetalk. Freetalk needs to work well. Any essential plugin changes for > Freetalk. (Mostly down to xor) > - The new wininstaller. This is essential, we need Vista support, > unfortunately it's not going to disappear much as we might want it to! (Easy) > - MHKs (or DHKs) - redundant CHKs for the top block of a splitfile (no, I > don't like RHK, it sounds too much like RSK, and R stands for Revocable not > Redundant). (Reasonably easy) > - Various minor usability tweaks. (Mostly easy but lots of it) > - Fix, or at least disappear, the "N peers forcibly disconnected due to not > acknowledging packets" bug/message. (???) > > Very important: > - Bloom filter sharing and all its dependancies: Client-cache, don't cache > while HTL is at maximum, transitional arrangements for old nodes' datastores. > (Estimate 4 weeks for working and thoroughly debugged) > > Fairly important: > - Multi-container freesite inserts. saces' work is pretty much working afaik, > but it needs to be made fully persistence aware and debugged and de-leaked. > - Update plugins over Freenet. > - Pause mode. > > Easy and would be nice: > - Solve the last-segment-has-less-redundancy problem for splitfiles. > - Guesstimate the monthly transfer usage for each bandwidth setting on the > bandwidth selection screen. > - Deploy the new x86-64 FEC code. > - IM-friendly noderefs. > > Would be nice: > - Javascript support for loading inline images on a page. > - Audio/video filters, plus an increased maximum file size when using the > progress page. > - Hierarchical DBRs for USK updating. > Loading official plugins off Freenet using the drop-down menu would be nice. I understand that this is not going to be in 0.8 because it requires official repositories on freenet with revocation key and everything, but IMHO it should have at least 'some' priority after 0.8 is released. Also, would it be possible to make an alternate update script that fetches the latest .jar off freenet instead of the website using that CLI tool mentioned in config page?
Re: [freenet-dev] What should be in 0.8, and when should it be released?
Matthew Toseland wrote: Ian has found some more funding, so we are okay for another 6-8 months. This will be announced properly shortly. This relieves the pressure to get something out regardless of what it is. So we can have a sensible discussion on: 1. What should be in 0.8? 2. When to release 0.8? My current view: We have a pretty good basis for a release already, but we need a lot more debugging, we probably need a load more optimisation on db4o (if discussions on the support list are anything to go by), there are a few features we have already agreed are essential, which are mostly easy, there is Freetalk, and there is Bloom filter sharing. IMHO Bloom filter sharing will make a *BIG* difference to Freenet's performance and therefore to the success (users, content, donations) of 0.8.0. We should aim for a beta in the not too distant future, however IMHO we should aim to implement the essential and very important features below before releasing it, and even a beta needs to be reasonably usable in terms of bugs; we will very likely need further optimisation on db4o to make Freenet practically usable for all users. So I am suggesting that we allow 2-3 months for feature work, fixing major bugs, and further optimising db4o, and release a beta in July. After that we have final bugfixing aiming for a final 0.8.0 before say October. FEATURES: Essential: - Freetalk. Freetalk needs to work well. Any essential plugin changes for Freetalk. (Mostly down to xor) - The new wininstaller. This is essential, we need Vista support, unfortunately it's not going to disappear much as we might want it to! (Easy) - MHKs (or DHKs) - redundant CHKs for the top block of a splitfile (no, I don't like RHK, it sounds too much like RSK, and R stands for Revocable not Redundant). (Reasonably easy) - Various minor usability tweaks. (Mostly easy but lots of it) - Fix, or at least disappear, the N peers forcibly disconnected due to not acknowledging packets bug/message. (???) Very important: - Bloom filter sharing and all its dependancies: Client-cache, don't cache while HTL is at maximum, transitional arrangements for old nodes' datastores. (Estimate 4 weeks for working and thoroughly debugged) Fairly important: - Multi-container freesite inserts. saces' work is pretty much working afaik, but it needs to be made fully persistence aware and debugged and de-leaked. - Update plugins over Freenet. - Pause mode. Easy and would be nice: - Solve the last-segment-has-less-redundancy problem for splitfiles. - Guesstimate the monthly transfer usage for each bandwidth setting on the bandwidth selection screen. - Deploy the new x86-64 FEC code. - IM-friendly noderefs. Would be nice: - Javascript support for loading inline images on a page. - Audio/video filters, plus an increased maximum file size when using the progress page. - Hierarchical DBRs for USK updating. Loading official plugins off Freenet using the drop-down menu would be nice. I understand that this is not going to be in 0.8 because it requires official repositories on freenet with revocation key and everything, but IMHO it should have at least 'some' priority after 0.8 is released. Also, would it be possible to make an alternate update script that fetches the latest .jar off freenet instead of the website using that CLI tool mentioned in config page? ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Usability test results
On Friday 15 May 2009 16:35:40 Thomas Sachau wrote: Matthew Toseland schrieb: On Thursday 14 May 2009 18:35:07 Thomas Sachau wrote: Matthew Toseland schrieb: My observation: Can we get rid of the I will configure it manually choice? And maybe the welcome page? (#3094) You want to force everyone to use the Wizard? Why would that be bad? What if i dont want to do use the Wizard? Also, if i removed the wizard done line (intentinally or by mistake), a new run would remove my custom settings. With the option, i can just stop the wizard and no harm done. If you know enough to skip the wizard you should shutdown the node, edit the config file and tell the node you have done the wizard! Related idea: We should maybe tell the user in the installer that they should use a separate browser for Freenet, rather than in the wizard? And then let them choose one, and then use it when they click on the icon to browse Freenet? (#3104) This would produce additional work for people packaging freenet, since they would have to warn the user themselves, while users tend to ignore the output of the package manager. So this would lower the chance of people noticing the request for a different freenet browser/profile and therefor i am against it. I suggest the current way: Warning during first call of the webinterface like it is currently done. Well, maybe on linux, with the packages that we don't have yet... Did you miss the Gentoo ebuilds? Isnt it a goal to get other distros to package it too? Just because it did not happen until now, doesnt mean it wont happen some time in the future. May just need more time since Gentoo as source based distro may be a bit better for packages than binary distros. No, it is a goal to package it with private repositories. Having a debian package that is frozen for 3 years is not useful at the present time. And if we have it for linux, why would you like to add additional code for windows (both in the installer and in freenet, which would have to detect the OS and then decide to show the warning or not)? Well, we could do something similar for *nix, no? Launch a suitable privacy enabled browser when the user runs the browse-freenet script? signature.asc Description: This is a digitally signed message part. ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] What should be in 0.8, and when should it be released?
On Saturday 16 May 2009 10:12:06 Luke771 wrote: Matthew Toseland wrote: Ian has found some more funding, so we are okay for another 6-8 months. This will be announced properly shortly. This relieves the pressure to get something out regardless of what it is. So we can have a sensible discussion on: 1. What should be in 0.8? 2. When to release 0.8? My current view: We have a pretty good basis for a release already, but we need a lot more debugging, we probably need a load more optimisation on db4o (if discussions on the support list are anything to go by), there are a few features we have already agreed are essential, which are mostly easy, there is Freetalk, and there is Bloom filter sharing. IMHO Bloom filter sharing will make a *BIG* difference to Freenet's performance and therefore to the success (users, content, donations) of 0.8.0. We should aim for a beta in the not too distant future, however IMHO we should aim to implement the essential and very important features below before releasing it, and even a beta needs to be reasonably usable in terms of bugs; we will very likely need further optimisation on db4o to make Freenet practically usable for all users. So I am suggesting that we allow 2-3 months for feature work, fixing major bugs, and further optimising db4o, and release a beta in July. After that we have final bugfixing aiming for a final 0.8.0 before say October. FEATURES: Essential: - Freetalk. Freetalk needs to work well. Any essential plugin changes for Freetalk. (Mostly down to xor) - The new wininstaller. This is essential, we need Vista support, unfortunately it's not going to disappear much as we might want it to! (Easy) - MHKs (or DHKs) - redundant CHKs for the top block of a splitfile (no, I don't like RHK, it sounds too much like RSK, and R stands for Revocable not Redundant). (Reasonably easy) - Various minor usability tweaks. (Mostly easy but lots of it) - Fix, or at least disappear, the N peers forcibly disconnected due to not acknowledging packets bug/message. (???) Very important: - Bloom filter sharing and all its dependancies: Client-cache, don't cache while HTL is at maximum, transitional arrangements for old nodes' datastores. (Estimate 4 weeks for working and thoroughly debugged) Fairly important: - Multi-container freesite inserts. saces' work is pretty much working afaik, but it needs to be made fully persistence aware and debugged and de-leaked. - Update plugins over Freenet. - Pause mode. Easy and would be nice: - Solve the last-segment-has-less-redundancy problem for splitfiles. - Guesstimate the monthly transfer usage for each bandwidth setting on the bandwidth selection screen. - Deploy the new x86-64 FEC code. - IM-friendly noderefs. Would be nice: - Javascript support for loading inline images on a page. - Audio/video filters, plus an increased maximum file size when using the progress page. - Hierarchical DBRs for USK updating. Loading official plugins off Freenet using the drop-down menu would be nice. I understand that this is not going to be in 0.8 because it requires official repositories on freenet with revocation key and everything, but IMHO it should have at least 'some' priority after 0.8 is released. Well, maybe for 0.8. The immediate future is 0.7.5. Also, would it be possible to make an alternate update script that fetches the latest .jar off freenet instead of the website using that CLI tool mentioned in config page? That would require a working node, in which case update over Freenet should work. signature.asc Description: This is a digitally signed message part. ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Usability test results
* xor x...@gmx.li [2009-05-15 18:24:53]: On Thursday 14 May 2009 19:35:07 Thomas Sachau wrote: Matthew Toseland schrieb: My observation: Can we get rid of the I will configure it manually choice? And maybe the welcome page? (#3094) You want to force everyone to use the Wizard? Because we were both on the same LAN, it did not connect, until I told him to set it to allow local addresses on that peer. There should be a checkbox when adding a noderef, defaulting to on, Friend may be on the same local network as me or something. (#3098) This is imho not usual, so i would set this to very low priority and only for advanced mode enabled. IMHO the current behavior is a bug though: We have the MDNS plugin, which is installed by default and should automatically (opennet-) connect nodes on a LAN. Did you even read the code? It was never intended to do that and doesn't. However it does not work because LAN connections are disallowed by default. IT WASN'T IMPLEMENTED AT ALL. signature.asc Description: Digital signature ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Usability test results
* Robert Hailey rob...@freenetproject.org [2009-05-15 13:04:58]: On May 15, 2009, at 11:57 AM, xor wrote: On Friday 15 May 2009 18:37:12 Thomas Sachau wrote: xor schrieb: On Thursday 14 May 2009 19:35:07 Thomas Sachau wrote: Matthew Toseland schrieb: My observation: Can we get rid of the I will configure it manually choice? And maybe the welcome page? (#3094) You want to force everyone to use the Wizard? Because we were both on the same LAN, it did not connect, until I told him to set it to allow local addresses on that peer. There should be a checkbox when adding a noderef, defaulting to on, Friend may be on the same local network as me or something. (#3098) This is imho not usual, so i would set this to very low priority and only for advanced mode enabled. IMHO the current behavior is a bug though: We have the MDNS plugin, which is installed by default and should automatically (opennet-) connect nodes on a LAN. However it does not work because LAN connections are disallowed by default. Afaik that plugin is no longer installed by default. Why not? I like the idea of self-organizing Freenet on LANs very much :) Especially if you consider large public or private LANs where two people running Freenet might not even know each others, so they couldn't get manually connected. Hmmm... but does a node want to advertise that we are running freenet? Maybe dependent on the security level setting. In either case, a small network would have major issues if it was not also connected to the large network, Ever heard of darknet? and so I think still that automatically detecting locality based on external ip address would help (e.g. a node could preferentially keep any opennet peers that it finds as being local). That's what darknet connections are about. Opennet nodes connecting to the same darknet nodes is a BUG which has already been reported and has a ticket on mantis. signature.asc Description: Digital signature ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Usability test results
* xor x...@gmx.li [2009-05-15 22:07:34]: On Friday 15 May 2009 20:04:58 Robert Hailey wrote: On May 15, 2009, at 11:57 AM, xor wrote: On Friday 15 May 2009 18:37:12 Thomas Sachau wrote: xor schrieb: On Thursday 14 May 2009 19:35:07 Thomas Sachau wrote: Matthew Toseland schrieb: My observation: Can we get rid of the I will configure it manually choice? And maybe the welcome page? (#3094) You want to force everyone to use the Wizard? Because we were both on the same LAN, it did not connect, until I told him to set it to allow local addresses on that peer. There should be a checkbox when adding a noderef, defaulting to on, Friend may be on the same local network as me or something. (#3098) This is imho not usual, so i would set this to very low priority and only for advanced mode enabled. IMHO the current behavior is a bug though: We have the MDNS plugin, which is installed by default and should automatically (opennet-) connect nodes on a LAN. However it does not work because LAN connections are disallowed by default. Afaik that plugin is no longer installed by default. Why not? I like the idea of self-organizing Freenet on LANs very much :) Especially if you consider large public or private LANs where two people running Freenet might not even know each others, so they couldn't get manually connected. Hmmm... but does a node want to advertise that we are running freenet? Maybe dependent on the security level setting. The more people confess that they are running Freenet, the more difficult it becomes to blame an individual as guilty for running it. Further, if you run opennet, you are screwed anyway, and MDNS is mostly useful for Opennet. In either case, a small network would have major issues if it was not also connected to the large network, and so I think still that automatically detecting locality based on external ip address would help (e.g. a node could preferentially keep any opennet peers that it finds as being local). Wouldn't it take much load off the internet, i.e. small bandwidth connections, if any nodes which are connected via LAN used the LAN for routing requests if possible? I think the LAN could behave like a single large node, i.e. requests for data which is stored within the LAN would be answered very fast, and much data could be stored if there are many nodes on a LAN. The idea is not new, at the beginning bloom-filter sharing was supposed to be for darknet nodes or local nodes. There is also a ticket on mantis describing that signature.asc Description: Digital signature ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Usability test results
Matthew Toseland schrieb: On Friday 15 May 2009 16:35:40 Thomas Sachau wrote: Matthew Toseland schrieb: On Thursday 14 May 2009 18:35:07 Thomas Sachau wrote: Matthew Toseland schrieb: My observation: Can we get rid of the I will configure it manually choice? And maybe the welcome page? (#3094) You want to force everyone to use the Wizard? Why would that be bad? What if i dont want to do use the Wizard? Also, if i removed the wizard done line (intentinally or by mistake), a new run would remove my custom settings. With the option, i can just stop the wizard and no harm done. If you know enough to skip the wizard you should shutdown the node, edit the config file and tell the node you have done the wizard! Is there a need for editing the config file? You can set everything with the config section too, but without the i want to do it myself, you cant disable the wizard from the GUI. Related idea: We should maybe tell the user in the installer that they should use a separate browser for Freenet, rather than in the wizard? And then let them choose one, and then use it when they click on the icon to browse Freenet? (#3104) This would produce additional work for people packaging freenet, since they would have to warn the user themselves, while users tend to ignore the output of the package manager. So this would lower the chance of people noticing the request for a different freenet browser/profile and therefor i am against it. I suggest the current way: Warning during first call of the webinterface like it is currently done. Well, maybe on linux, with the packages that we don't have yet... Did you miss the Gentoo ebuilds? Isnt it a goal to get other distros to package it too? Just because it did not happen until now, doesnt mean it wont happen some time in the future. May just need more time since Gentoo as source based distro may be a bit better for packages than binary distros. No, it is a goal to package it with private repositories. Having a debian package that is frozen for 3 years is not useful at the present time. And if we have it for linux, why would you like to add additional code for windows (both in the installer and in freenet, which would have to detect the OS and then decide to show the warning or not)? Well, we could do something similar for *nix, no? Launch a suitable privacy enabled browser when the user runs the browse-freenet script? You dont know the user system. While windows user systems may be similar to each others, this is not true for linux. Where would you place that script? How would you check which browser the user wants to use? This idea looks more like the way user handling is done on windows or ubuntu: Expect him to know nothing and try to do everything for him. Might be nice for beginners and if it works, but makes things worse for experienced users, who want to do it different and also makes it harder, if there are problems. Imho you cant beat stupidity. Either users read a message and act the right way or they dont. You cannot prevent them from doing bad things. Additionally, Gentoo is about choice, if there is a warning, the user can choose, with a forcing script, there is no choice, which is a bad idea for this philosophy, therefor i vote against such a script for linux. signature.asc Description: OpenPGP digital signature ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Usability test results
On Saturday 16 May 2009 15:02:19 Thomas Sachau wrote: Matthew Toseland schrieb: On Friday 15 May 2009 16:35:40 Thomas Sachau wrote: Matthew Toseland schrieb: On Thursday 14 May 2009 18:35:07 Thomas Sachau wrote: Matthew Toseland schrieb: My observation: Can we get rid of the I will configure it manually choice? And maybe the welcome page? (#3094) You want to force everyone to use the Wizard? Why would that be bad? What if i dont want to do use the Wizard? Also, if i removed the wizard done line (intentinally or by mistake), a new run would remove my custom settings. With the option, i can just stop the wizard and no harm done. If you know enough to skip the wizard you should shutdown the node, edit the config file and tell the node you have done the wizard! Is there a need for editing the config file? You can set everything with the config section too, but without the i want to do it myself, you cant disable the wizard from the GUI. Related idea: We should maybe tell the user in the installer that they should use a separate browser for Freenet, rather than in the wizard? And then let them choose one, and then use it when they click on the icon to browse Freenet? (#3104) This would produce additional work for people packaging freenet, since they would have to warn the user themselves, while users tend to ignore the output of the package manager. So this would lower the chance of people noticing the request for a different freenet browser/profile and therefor i am against it. I suggest the current way: Warning during first call of the webinterface like it is currently done. Well, maybe on linux, with the packages that we don't have yet... Did you miss the Gentoo ebuilds? Isnt it a goal to get other distros to package it too? Just because it did not happen until now, doesnt mean it wont happen some time in the future. May just need more time since Gentoo as source based distro may be a bit better for packages than binary distros. No, it is a goal to package it with private repositories. Having a debian package that is frozen for 3 years is not useful at the present time. And if we have it for linux, why would you like to add additional code for windows (both in the installer and in freenet, which would have to detect the OS and then decide to show the warning or not)? Well, we could do something similar for *nix, no? Launch a suitable privacy enabled browser when the user runs the browse-freenet script? You dont know the user system. While windows user systems may be similar to each others, this is not true for linux. Where would you place that script? How would you check which browser the user wants to use? This idea looks more like the way user handling is done on windows or ubuntu: Expect him to know nothing and try to do everything for him. Might be nice for beginners and if it works, but makes things worse for experienced users, who want to do it different and also makes it harder, if there are problems. Imho you cant beat stupidity. Either users read a message and act the right way or they dont. You cannot prevent them from doing bad things. Additionally, Gentoo is about choice, if there is a warning, the user can choose, with a forcing script, there is no choice, which is a bad idea for this philosophy, therefor i vote against such a script for linux. Well, we already have a Browse Freenet script on all three platforms. Currently it detects browsers that we know about. You don't have to use it if you don't want to. But we should extend it to use incognito mode if possible, and to favour browsers with such support. I dunno how we can determine whether such a mode works with the particular installed version though... signature.asc Description: This is a digitally signed message part. ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Infocalypse feedback (Mercurial over Freenet)
On Monday, 11. May 2009 21:20:49 Arne Babenhauserheide wrote: The only problem I still have is that keeping the uris in the central config file didn't work (all paths in the config file were lowercase while the real paths aren't - maybe that's connected to the issue). This problem seems fixed by now; works for me on amd64, Gentoo GNU/Linux - thanks to feral code wright! Information: $ hg version Mercurial Distributed SCM (version 437e06bbd11e+20090515) # means: current hg head from the main development repo $ python --version Python 2.5.4 Best wishes, Arne --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- - singing a part of the history of free software - http://infinite-hands.draketo.de signature.asc Description: This is a digitally signed message part. ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Usability test results
On Friday, 15. May 2009 22:07:34 xor wrote: Wouldn't it take much load off the internet, i.e. small bandwidth connections, if any nodes which are connected via LAN used the LAN for routing requests if possible? I assume that it would also help privacy, because then timing analysis and similar would become much harder, since external nodes can't look into the LAN. To avoid too easy internal traceability, the option of switching to port 8080 or 5223 (jabber server with SSL) (or 80 if started with sufficient rights) would be nice. It would avoid the question what exactly are you running? Best wishes, Arne --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- - singing a part of the history of free software - http://infinite-hands.draketo.de signature.asc Description: This is a digitally signed message part. ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Usability test results
On Saturday 16 May 2009 23:20:08 Arne Babenhauserheide wrote: On Friday, 15. May 2009 22:07:34 xor wrote: Wouldn't it take much load off the internet, i.e. small bandwidth connections, if any nodes which are connected via LAN used the LAN for routing requests if possible? I assume that it would also help privacy, because then timing analysis and similar would become much harder, since external nodes can't look into the LAN. To avoid too easy internal traceability, the option of switching to port 8080 or 5223 (jabber server with SSL) (or 80 if started with sufficient rights) would be nice. It would avoid the question what exactly are you running? Not much point hiding it if you're broadcasting the existence of nodes via MDNSDiscovery! signature.asc Description: This is a digitally signed message part. ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Please help us test the new wininstaller (Vista users especially welcome)
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Using the current wininstaller (FreenetInstaller-1210), I installed into Windows 7 x64 Build 7100. Installed java as prompted (x32). Install went very well. The uninstaller leaves user 'Freenet' and I updated my open ticket(bug) with these results. Freenet found that I had Chrome installed and used it, even though it was not my default browser.Bug/Feature? -BEGIN PGP SIGNATURE- Charset: UTF8 Note: This signature can be verified at https://www.hushtools.com/verify Version: Hush 3.0 wkYEARECAAYFAkoPUi8ACgkQDAg0OvA3V4BvTACfctkSvN531D0WN5mDSyFd1bcTCHMA n0Fmqx36EEj+tFzptUzTxsAYgtwX =Wrc+ -END PGP SIGNATURE- -- Free information on a POS system that meets your needs. Click here. http://tagline.hushmail.com/fc/BLSrjkqZugEikWAyHtCFVznVXfLKxokMTdj0gKb01Zq8iQxm3rynr2M6HnC/ ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
Re: [freenet-dev] Wininstaller deployed
On Thu, May 14, 2009 at 1:31 PM, Zero3 ze...@zerosplayground.dk wrote: Matthew Toseland skrev: We now include wget.exe and sha1test.jar. Also, I have put the update.cmd in update-new.cmd on emu and updated it to fetch itself, and made it use icacls on win 5.2 (XP64, win2k3 server etc). And fixed some problems with its use of the start/stop scripts, and a bug that was causing it not to switch between builds that had already been downloaded (which still exists in the other version of update.cmd iirc). So it works now, despite java not being on the path (causing the verification not to run; why doesn't it just fail?). Regarding these comments in update.cmd: :Assume that it was running, no way to easily tell - FIXME what to grep for in the service list when multiple installs? and :: FIXME do we need a new error handling section for the new .exe? Will it handle errors itself? The service name is freenetinstallsuffix, where installsuffix is the contents of installid.dat in the install dir (empty on first install, _2 on second install, _3 on third install and so on). Run start.exe /? and stop.exe /? to see command line options and return codes (or look in the source: src_freenethelpers/FreenetStart.ahk and src_freenethelpers/FreenetStop.ahk). - Zero3 I have committed a new version of the script that sets restart flag on exit code 0, errors on 1 and continues on 2 (node already stopped). Zero3- Do the start/stop exe's look for the installid.dat and use that when starting and stopping the node? I assume so, and that it is a relative path .ie. ..\installid.dat If so then we don't need to grep for the service in the script at all? ___ Devl mailing list Devl@freenetproject.org http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
